Dear Edward,
> The fMRI model that I am attempting to create is based upon the
> following experimental design: TR = 2; total images = 105; 2 trial
> blocks of 35 images each; Block 1 prefaced by roughly 13-14 scans of
> rest; Block 2 prefaced by roughly 13-14 scans of rest, and followed by
> roughly 8 scans of rest. (I say roughly because each subjects model is
> different) Anyhow, I'm not clear as to how I should model the rest
> period and how SOA relates to it. Do I use SOA to model rest, or
> rather the transition from rest to task within a session. I suppose my
> basic question concerns the application of SOA into the model and the
> correct modeling method for rest periods, be they before or after a
> task block.
Assuming your block length is constant (i.e. only the rest periods are
variable) simply specify one trial type and enter the SOAs (in SPM99)
as the onsets of the test blocks. The intervening rest periods will be
modelled implicitly by the constant and other drift terms, allowing you
to focus on the excurions of signal incurred by the task from this
baseline.
I hope this helps - Karl
|